diffstride
diffstride copied to clipboard
Why stop_gradient operator is used in diffstride?
Firstly, I want to appreciate for sharing great work with other researchers.
I write this issue because I have a question about using the stop_gradient operator.
When I saw other usages of the stop_gradient operator, they were usually used to prevent the gradient from flowing to the graph.
However, I think that the stop_gradient operator might be unnecessarily used in diffstride because the cropping operation is originally not differentiable.
Do I understand it correctly? If I misunderstand it, please give me some advice about it.
Thank you.
Cropping is a discrete function, which inherently makes it non-differentiable.